The Investment Behavior of Private Equity Fund Managers * †

Alexander Ljungqvist Matthew Richardson Stern School of Business Stern School of Business New York University New York University and CEPR and NBER

First draft: June 15, 2003 This draft: October 8, 2003

* We are grateful to an anonymous institutional investor for making the data used in this study available, to the Salomon Center at NYU Stern for generous financial assistance, and to Eric Green for many helpful discussions and suggestions. We also thank Colin Blaydon, Wayne Ferson, Steve Kaplan, Holger Mueller, Maureen O’Hara, Antoinette Schoar, Robert Whitelaw, Jeff Wurgler, participants at the 2003 Western Finance Association meeting, the 2003 Stanford GSB/NYSE Conference, the NYU Monday seminar series, and numerous members of the private equity community for many helpful comments. We are grateful to Eric Stern for excellent research assistance. All errors are our own. † Address for correspondence: Salomon Center, Stern School of Business, New York University, Suite 9-160, 44 West Fourth Street, New York NY 10012-1126. Fax 212-995-4220. e-mail aljungqv@stern.nyu.edu (A. Ljungqvist), mrichar0@stern.nyu.edu (M. Richardson).

The Investment Behavior of Private Equity Fund Managers

Abstract Using a unique dataset of private equity funds over the last two decades, this paper analyzes the investment behavior of private equity fund managers. Specifically, we investigate the determinants of how quickly funds invest their capital and exit their portfolio companies, and what determines the success of their investments. Controlling for fund characteristics and market conditions, we show that the competitive environment facing fund managers affects how investments are managed. In particular, we show that time variation in the availability of investment opportunities and competition for deal flow with other private equity funds differentially affect the time a fund takes to invest its omitted capital, the time to return this capital, and the resulting success of these investments. These findings provide complementary evidence to recent papers documenting the determinants of fund performance. JEL Classification: G23, G11. Key words: Private equity; Venture capital; Alternative investments; Fund management.

1. Introduction

The goal of this paper is to try and better understand how private equity funds (“PEF”s) make

investment decisions in a competitive market. How competition affects the private equity sector is an

interesting question because it brings together the interaction of three distinct economic agents –

investors, financial intermediaries (i.e., PEFs) and entrepreneurs. Our paper complements a recent

literature that looks at this question theoretically, most notably Inderst and Mueller (2003). With

specific reference to PEFs, Inderst and Mueller show how capital market conditions affect (i) the

valuation and success of PEFs’ investments, and (ii) the search time and screening by PEFs with

respect to their investments. In this paper, we investigate some of these issues empirically by

documenting the microeconomics of how PEFs invest their capital and exit their portfolio investments

when faced with different economic environments.

Specifically, we analyze the role of competition among PEFs and the stickiness with which the

PEF market adjusts to demand shocks in the context of four questions: What determines (i) the speed

with which PEFs invest their capital over time, (ii) how long it takes them to return capital to their

investors, (iii) when they exit their portfolio investments, and (iv) what returns they earn on their

portfolio companies? For this, we make use of a unique and proprietary dataset made available to us by

one of the largest institutional investors in private equity. Our dataset includes, among other items,

precisely dated cash flows representing investments in 3,800 portfolio companies by several hundred

private equity funds. The dataset accounts for approximately 20% of all capital raised by PEFs over the

period 1981 to 2001 and so affords a comprehensive view of investment behavior in the private equity

fund industry.

Our dataset has two important advantages over others used in the literature. First, because we know

the exact timing of the cash flows (and thus the timing of both the investment and exit decision), we

are able to relate PEF managers’ decisions to measures of market competition and investment

opportunities that are distinct from each other. Employing these measures, we find evidence consistent

with the importance of changes in the demand for private equity capital and stickiness in its supply for

explaining the observed behavior of PEF managers. For example, we show that time variation in the

availability of investment opportunities and competition for deal flow with other private equity funds

significantly affect the time a fund takes to invest its committed capital and then return it to its

investors.

The second advantage is that we are able to document not only aggregate PEF behavior at the fund

level, but also the individual investment decisions within a PEF’s portfolio. Therefore, we are able to

2

analyze the determinants of how long it takes PEFs to exit their investments, that is, the length of the

“holding period” for each portfolio company. Similarly, we analyze the multiples on investment that

PEFs earn on each of their portfolio holdings. We show that holdings periods are shorter and the

corresponding success rates are higher following improvements in the availability of investment

opportunities. Analogously, investments are held for longer, and are less successful, when competition

for deal flow is tougher. We argue that this is consistent with an underlying model for imperfect

competition for PEFs, and that it relates closely to implications from Inderst and Mueller (2003).

Our paper adds to several well-documented empirical facts from an emerging literature on private

equity that begins to piece together the interaction of PEFs with the rest of the economy. First,

Gompers and Lerner (1998) document that aggregate flows into the private equity sector tend to be

driven by demand shocks. Second, somewhat paradoxically, Gompers and Lerner (2000) and Kaplan

and Stein (1993) argue and document that too much capital can flow into the PEF sector, leading to the

“money chasing deals” phenomenon. Third, Kaplan and Schoar (2003) document that better

performing funds have an easier time raising capital for follow-on funds and that these funds perform

better than the overall pool of funds. We argue that our findings are consistent with these facts in the

context of a model for imperfect competition among PEFs.

We organize the paper as follows. In Section 2, we describe a framework for understanding

competition in the PEF sector and discuss several implications of the model. Because the dataset is

new, Section 3 describes in detail its various properties. Of some interest, we provide a comparison of

our sample of PEFs to the larger (albeit much less detailed) Venture Economics dataset used by other

authors. Sections 4 and 5 provide the core results of the paper by documenting the investment patterns

of private equity funds over the last 20 years. In Section 4, we investigate the key determinants of the

draw down and exit decisions of PEFs at the fund level. Section 5 provides results on PEFs’ exit

strategies at the individual investment level and documents their “hit” rates with a corresponding

analysis of the determinants. Section 6 suggests a potential important area of future research that ties

into ours and existing results in the PEF literature.

2. Framework: The Competitive Market for PEFs

Consider the market for private equity. PEFs raise money from institutional and other investors and

channel it to entrepreneurs. PEFs are typically structured as limited partnerships with a fixed (usually

ten-year) life and thus resemble closed-end funds. The so called general partners or GPs managing the

fund receive an annual fee of around 1-2% of capital under management and take a slice of the fund’s

profits (the carried interest or carry), typically 20%. Investors (the limited partners or LPs) commit

3

capital to the fund which GPs draw down over the fund’s life whenever they wish to invest in a

portfolio company.

In this paper, we assume that there is competition (albeit perhaps imperfect) among (i) the

institutions and individuals who wish to invest in PEFs, and (ii) the PEF managers who wish to invest

in particular entrepreneurial projects. Positive net present value projects correspond to business plans

that produce competitive advantage. Though entrepreneurs eventually face competition, they are at

least initially monopolists with respect to their business plans. For the moment, we assume the

entrepreneurs capture all the rents their projects produce. In this type of market setting, we are

interested in the determinants of the investment decisions of private equity fund managers.

Assuming this asset market is rational and competitive, investors will provide capital to private

equity funds until their risk-adjusted expected returns (net of fees) equal the expected returns they

could earn elsewhere. Suppose a technological shock hits the entrepreneurial market. This shock could

be the development of the personal computer, changes in the way the FDA approves drugs, the

development of the internet, the creation of the high-yield debt market, and so forth. Conditional on the

shock, entrepreneurs demand capital from the market. The literature argues that private equity funds

are the cheapest source of financing when private firms are subject to extreme informational

asymmetries and high degrees of uncertainty (e.g., Gompers and Lerner (1999b)). In a perfect world,

capital flows immediately into PEFs which in turn provide capital to the entrepreneurs. This story of

capital flows is consistent with Gompers and Lerner (1998) who argue that most flows into and out of

this sector are driven by demand shocks.

To the extent that there are net present value gains, the excess returns are assumed to accrue to the

entrepreneur. Presumably, the services provided by the PEFs are fully compensated for by the stakes

they take in the ventures, which in turn are offset by the fees paid to the PEFs by the investors.1 (See,

for example, Gorman and Sahlman (1989), Palepu (1990), Gompers and Lerner (1999b), and

Hellmann and Puri (2002) for the types of specialized services that PEFs provide to entrepreneurs.)

There are strong reasons, however, to question this view of the PEF market. In the Inderst and

Mueller (2003) model, the entrepreneur’s bargaining power relative to the venture capitalist varies with

changes in the demand and supply of capital. If the supply of PEFs in the short run is somewhat fixed,

then a sudden shift in the demand for entrepreneurial capital will lead to a transfer of rents from the

1 Jones and Rhodes-Kropf (2003) argue that, due to the principal-agent problems associated with private equity investing, PEFs necessarily hold undiversified positions. Thus, part of the compensation to PEFs relates to the level of idiosyncratic risk faced by fund managers.

4

entrepreneur to existing PEFs (and their investors2) until the supply of PEFs catches up (see also

Sahlman (1990)).

The stickiness of the PEF market is not without justification. First, relative to other asset classes, it

is well known that private equity investments are illiquid. That is, there is no active secondary market

for such investments, investors have little control over how the capital is invested, and the investment

profile covers a long horizon. If the supply of available capital that puts zero price on liquidity is

limited, then this will lead to rent transfers from the entrepreneur.3 Second, once a fund has been

raised, its size cannot subsequently be increased (though in recent years some funds have cut their

size). Thus, reacting to a shift in demand requires raising a new fund which at minimum takes several

months. Third, and perhaps more important, it is often argued that PEF managers possess unique skills

that are not easily duplicated overnight. This limits established GPs’ ability to raise additional funds (to

avoid overstretch) and constrains to some extent entry by new fund managers. The skills in question

include the ability to screen investment proposals and monitor entrepreneurs (both indirectly and

directly through sitting on companies’ boards), a “rolodex” of contacts who can help add value to the

ventures, and access to financing (e.g., Gompers (1995), Gompers and Lerner (1996), Lerner (1994),

and Hellmann and Puri (2002)). The contacts in particular are built up through years of experience and

working in the industry.

Some argue that there is an additional imperfection in the PEF market. At times, too much capital

flows into the PEF sector, so that capital investment can actually overshoot, leading to the “money

chasing deals” phenomenon documented by Gompers and Lerner (2000) and Kaplan and Stein (1993),

and studied in a specific example by Sahlman and Stevenson (1986). This apparent breakdown of

efficiency on the investor side is usually considered behavioral (see, for example, the herding literature

and, in particular, Wermers (1999) for his application to mutual funds). However, it could simply

reflect investors trying to take advantage of the stickiness of the PEF market, which makes excess

returns possible. If the technological shock is unpredictable, then the returns earned in “sticky” markets

and in “money chasing deals” markets may average out to be normal rates of return. This is an

important area of research which we discuss in greater detail in the paper’s conclusion.

2 Who ultimately earns the excess rents would depend on the contractual arrangements between the PEF and the investors. To the extent that there is little variation across contracts, investors earn some of the excess returns. Excess returns may be offset by poor returns if the investor mis-times the cycle (see below). Alternatively, it has been argued that there exist subtle, yet important, differences across PEF contracts (e.g., Gompers and Lerner (1999a)). 3 Recent work by Lerner and Schoar (2003) argues that incentive problems between PEFs and investors can be alleviated by the PEF using illiquidity to screen for investors who are less subject to liquidity shocks. For our example, the PEFs would need to trade off the benefits of having liquid investors versus the shortage of such investors.

5

Of course, these two views of the PEF market are entirely mutually consistent. A technological

shock hits the entrepreneurial market, leading to an initial shortage of PEF-based capital and high

returns, only to be supplemented (perhaps overly so) with new PEF capital once the market adjusts.

Note that this is also a direct result of equilibrium in the Inderst and Mueller (2003) model.

Our framework has implications both for how investment decisions are made and for their relative

success. Consider a fund manager’s investment behavior following a technological shock. Ceteris

paribus, the manager of a fund that is already in place should invest his capital as fast as possible in

promising projects, before new PEFs are created to invest in the same opportunities. Thus a PEF’s

investment rate should increase as more promising investment opportunities arise. These investments

should also yield higher returns. On the other hand, holding the number of projects fixed, an increase

in competition for deal flow among the PEFs makes it harder for the fund manager to find the so-called

“diamonds in the rough”, leading to a slow-down in the investment rate. This effect corresponds to

Inderst and Mueller’s (2003) prediction that the PEF manager’s search time increases when

competition intensifies. Greater competition presumably also implies that the fund manager will find it

more difficult to extract rents from the entrepreneur. A manager trying to maximize the returns on the

fund’s investments will then take longer to invest the fund’s capital to avoid overpaying.

In terms of capital return and exit decisions, we expect that funds that faced tough competition

when making their investments will take longer to return capital to their investors and exit their

portfolio holdings later. In part this follows because, as mentioned above, funds take longer to invest

when competition is tough. Moreover, we expect funds facing tough competition to make more

marginal investments which need more “nursing” before they can be exited, and which arguably have

higher mortality rates (see Bengtsson, Kaplan, Martel, and Strömberg (2002) for evidence that VCs

screen less when competition for deal flow is intense). Improvements in the investment environment

(say in response to technological shocks), on the other hand, should serve to accelerate both capital

returns at the fund level and exits at the investment level.

A corollary of this analysis is that PEFs that can react quickly to market conditions, say by being

able to raise funds on short notice, would have a comparative advantage. Along these lines, Kaplan and

Schoar (2003) find that better performing funds have an easier time raising capital for follow-on funds.

Interestingly, their comprehensive analysis of the performance of 746 PEFs raised between 1980 and

1995 shows that (i) follow-on funds perform better than the overall pool of funds, and (ii) funds raised

in boom times (i.e., with considerable PEF competition) tend to perform worse. These results are

consistent with the notion of a competitive advantage and show support for the model of the private

6

equity market described above. We will return to the implications of our analysis for who earns excess

returns in the private equity market in the concluding section.

3. Sample and Data

3.1 Overview of Dataset

Our dataset is derived from the records of one of the largest institutional investors in private equity

in the U.S. We will refer to this investor as the “Limited Partner”. As a condition for obtaining the

data, we have agreed to identify neither the Limited Partner nor the names of the funds or portfolio

companies in the dataset. The Limited Partner began investing in private equity in 1981, in the wake of

the institutionalization of the private equity industry following the 1980 ERISA “Safe Harbor”

regulation, and has since invested in hundreds of funds, all of which are included in our analysis.4 The

funds, in turn, have invested in 3,800 portfolio companies. The number of funds the Limited Partner

participated in increased throughout the 1990s, peaking in 1999-2000, similar to the pattern

documented for PEFs in general by Venture Economics (VE), a commercial data vendor.

Table 1 presents descriptive statistics for our sample. To protect the Limited Partner’s identity, we

have agreed not to disclose in this table certain characteristics of funds raised after 1993, such as their

number and average size, as these are still active investments. (However, we include the underlying

cash flow data for all funds in our subsequent analyses.) The table thus contains more complete

information for the 73 private equity funds the Limited Partner invested in between 1981 and 1993.

We define these funds as “mature” funds since they are around ten or more years old and have

completed their investment activity and capital distributions.

Our dataset contains both venture capital and buyout funds.5 For the entire period from 1981 to

2001, a quarter of funds, representing 14.8% of fund capital, are venture funds. This differs from the

more comprehensive sample of funds tracked by VE, where venture funds account for 41.5% by

capital. Our Limited Partner thus invests disproportionately in buyout funds.

In the private equity industry, fund size is usually expressed as the sum of investors’ “capital

commitments”. The capital commitment is the maximum amount of money an investor can be asked to

contribute over the life of the fund. Between 1981 and 2001, sample funds had aggregate commitments 4 The institutionalization of the private equity industry is commonly dated to three events: the 1978 Employee Retirement Income Security Act (ERISA) whose “Prudent Man” rule allowed pension funds to invest in higher-risk asset classes; the 1980 Small Business Investment Act which redefined PEF managers as business development companies rather than investment advisers, so lowering their regulatory burdens; and the 1980 ERISA “Safe Harbor” regulation which sanctioned limited partnerships which are now the dominant organizational form in the industry. 5 Venture funds are those identified as “Venture Capital” by Venture Economics. Most non-venture funds are flagged as “Buyout” (90.4%); the remainder are flagged as “Generalist Private Equity” (3.8%), “Mezzanine” (4.8%), and “Other Private Equity” (1%). We will refer to these funds collectively as buyout funds.

7

of $207 billion (in nominal terms). Mature funds had aggregate commitments of $36.7 billion, with the

average fund raising $502.8 million. Buyout funds were substantially larger than venture funds,

averaging capital commitments of $599.7 million versus $227.5 million. Compared to the sample of

funds tracked by VE, our funds are large: Kaplan and Schoar (2003) report average fund sizes for

buyout and venture funds of $262 million and $53 million between 1980 and 1995, respectively.

Our Limited Partner’s investment in the private equity industry is sizeable. Between 1981 and

2001, it committed $5.5 billion to PEFs, with the median fund receiving $10 million. As a fraction of

total fund size, the Limited Partner committed 4.7% of the average fund’s capital, making it one of the

larger investors.

3.2 Sample Selection

Apart from being skewed toward larger and buyout funds, how representative is our funds sample?

First, note that our data are not subject to survivorship bias as all investments the Limited Partner has

made since 1981 are included. Second, our sample covers a large fraction of the PEF “universe”. The

$207 billion raised in aggregate by our funds represent 17.5% of the $1.184 trillion in aggregate

commitments in the broader VE sample over the 1981-2001 period (see Table 1). Our coverage is even

better among buyout funds, accounting for 29.3% of capital committed to those funds. Thus, our

sample represents a reasonable cross-section of large buyout funds and a smaller cross-section of large

venture funds. By implication, our results may not be representative of the investment behavior of

smaller funds.

Third, the extent to which the funds in our dataset are representative of the universe of private

equity funds depends in part on the Limited Partner’s investment strategy. For instance, it would be

problematic if the Limited Partner only invested in follow-on funds raised by managers with proven

track records, in the manner of a fund-picking “fund-of-funds” operation. This is not the case. Table 1

shows that in our dataset, 28.9% of funds raised between 1981 and 2001 are first-time funds, 20.6%

are second funds, 11.6% are third funds, and the remaining 39.0% are later funds. Among mature

funds raised before 1994, as many as 34.8% are first-time funds, a rate that is not much lower than the

40% reported by Kaplan and Schoar (2003) for the VE database.

In part, the relatively high incidence of first-time funds follows from the Limited Partner’s twin

investment objectives: not only to obtain the highest risk-adjusted return, but also to increase the

likelihood that the funds will “purchase” the services our Limited Partner’s corporate parent has to

offer. Economies of scale in the provision of these services explain our Limited Partner’s tendency to

invest in larger than average funds. These services are arguably more attractive to first-time funds that

8

have yet to build up relationships.

Fourth, as in all studies with limited samples, the question of selection bias arises. There are two

possibilities here. The first is that the Limited Partner picked PEFs which were ex post unusual in how

they invested and distributed capital. For example, with respect to capital returns, perhaps the Limited

Partner chose more liquid investments (i.e., PEFs that paid off more quickly) or had extraordinary

fund-picking ability in choosing PEFs that ended up with many more “hits”. We tend to discount this

possibility. As described above, the Limited Partner’s primary motivation for investing in these funds

was to build relationships for the benefit of its corporate parent. Moreover, we know that the Limited

Partner is not organized as a professional fund-picking (“fund-of-funds”) operation. Finally, members

of the private equity community who have seen our results tell us they look representative.

The second possibility is that the Limited Partner might be exceptional in that it “survived” these

past 20 years, so that we observe its data more by virtue of its luck in investing in winner funds than

because private equity funds were good investments on average. While this point is probably not

particularly relevant (investing in private equity accounts only for a small part of the Limited Partner’s

overall business), we can shed more light on it directly by comparing the performance of our funds to

the performance of the wider VE sample.6 Kaplan and Schoar (2003) report that cash flow IRRs

averaged 18% among the 746 mature funds raised in 1980-1995 that are covered by VE. In our sample

of 73 mature (albeit larger) funds, IRRs average 18.13%, which is unlikely to be significantly different.

3.3 Cash Flows

The Limited Partner made available to us the complete cash flow records for all its private equity

investments up to May 31, 2001. We subsequently obtained additional data up to September 30, 2002

for 21 funds that were close to maturity, thus increasing the number of funds that have been liquidated

or are close to liquidation. A typical record consists of the date and amount of the cash flow, the fund

and portfolio company to which it relates, and the type of transaction. Transaction types include

“disbursements” (investments in portfolio companies) and “exits” (receipt of cash inflows from IPOs

or trade sales); dividends or interest paid by portfolio companies; annual management fees (typically 1-

2% of committed capital); and (occasional) interest payments on cash held by GPs prior to making an

investment. The data do not separately record the GPs’ share in a fund’s capital gains (usually 20%), as

GPs transmit capital gains to investors net of their “carried interest”.

The cash flows involve four types of investment scenarios. 1) Cancelled transactions: a cash call

6 We thank Steve Kaplan for this suggestion.

9

followed shortly after by the return of the cash, along with bank interest. 2) Write-offs: cash outflow(s)

without subsequent cash inflow, or with a subsequent accounting (non-cash) entry flagging a “capital

loss”. 3) Cash distributions following successful exits (in the form of an IPO or a trade sale): cash

outflow(s) followed by cash inflow(s). And 4), stock distributions following successful exits: cash

outflow(s) followed by a non-cash entry reflecting receipt of common stock. The stock would be the

portfolio company’s in the case of an IPO, and the buyer’s in the case of a sale to a publicly traded

firm. Following a stock distribution, one of two things can happen: the Limited Partner sells the stock,

or it holds it in inventory. Sales show up as cash inflows. Positions that are held in inventory are

marked to market periodically (usually monthly), but they are obviously not cash. Upon receipt of

distributed stock, our Limited Partner virtually always liquidates the distributed stock.

3.4 Portfolio Compositions and Industry Specializations

Venture Economics assigns companies to six broad industry groups: “Biotechnology”,

“Communications and Media”, “Computer Related”, “Medical/Health/Life Science”,

“Semiconductors/Other Electronics”, and “Non-High-Technology”. Companies that do not appear in

VE are assigned manually to these industry groups, using Dun & Bradstreet’s Million Dollar

Database, SIC codes that are available from standard sources for companies that have gone public,

verbal information contained in fund reports received by our Limited Partner, and news and web

searches. 209 companies that cannot be assigned unambiguously to one of the six VE groups are

assigned to a new “Miscellaneous” group.

Of the 3,800 companies that our sample funds invested in between 1981 and 2001, 3% are assigned

to “Biotechnology”, 17% to “Communications and Media”, 18% to “Computer Related”, 7% to

“Medical/Health/Life Science”, 4% to “Semiconductors/Other Electronics”, 45% to “Non-High-

Technology”, and 6% to “Miscellaneous”. The high proportion of non-high-technology portfolio

companies reflects the large number of buyout funds in the sample.

Funds rarely invest in only one industry. We take a sample fund’s industry specialization to be the

broad VE industry group that accounts for most of its invested capital. On this basis, 14% of funds

specialize in “Communications and Media”, 11% in “Computer Related” companies, 4% in

“Medical/Health/Life Science”, 3% in “Semiconductors/Other Electronics”, and 59% in “Non-High-

Technology”. Our sample contains no funds specializing in “Biotechnology”.

4. The Investment and Capital Return Decisions of Private Equity Funds

There is a large empirical literature on the investment process of private equity funds. However,

this literature almost exclusively analyzes the contractual relations between PEFs and the firms in their

10

portfolios. (See Gompers (1995), Lerner (1994), Kaplan and Strömberg (2003), and Hellmann and Puri

(2002), among others.) In this section, we take contracts as given and instead empirically analyze how

a private equity fund invests its capital over its life in the context of the descriptive model of the PEF

market outlined in Section 2. When a PEF receives a capital commitment from investors, the capital is

not put to use immediately and instead is drawn down only when the PEF is ready to invest in a

portfolio company. We document the dynamics of these draw downs over a fund’s life, as well as how

quickly capital gets returned. Of particular importance, we show that there is substantial cross-

sectional variation in draw down rates and capital return rates, and we perform a duration analysis of

the determinants of how fast or how slowly these flows occur. However, we first document some new

stylized facts that serve as a backdrop for our analysis.

4.1 Cash Flow Patterns: Draw Downs and Capital Distributions

Table 2A shows how much of the committed capital was drawn down by the earlier of the end of

our sampling period or a fund’s liquidation date. The average fund in our sample has drawn down

67.3% of committed capital. However, this understates draw downs as the more recent funds are not

yet fully invested. The 73 funds raised between 1981 and 1993 invested on average 94.8% of

committed capital. Average draw downs are around 90% of committed capital for funds raised up to

1996, with later vintages still actively investing and so still in what is called the “commitment period”.

It is arguable when a fund is fully invested. Among the funds raised between 1981 and 1993 that

have subsequently been liquidated, some never invested more than 60 to 70% of committed capital. In

the overall 1981-2001 dataset, 55.6% of funds have invested at least 70% of committed capital, and

49.5% have invested 80% or more as of the end of our sampling period. These might reasonably be

thought of as fully, or close to fully, invested. They include a few recent funds that invested their

committed capital very rapidly: 40% of the 1998 vintage funds and 10% of the 1999 vintage funds had

already invested at least 70% of committed capital by May 2001.

Private equity funds rarely draw down their committed capital at the outset, issuing capital calls

instead when investment opportunities present themselves. Figure 1 sheds light on the time profile of

draw downs for the average sample fund. The figure shows average cumulative draw downs for each

year of a fund’s life (counted from 1 to 10), divided by committed capital. The average fund draws

down 16.28, 20.35, and 20.15% of committed capital in its first three years of operation, so it is 56.8%

invested by the end of year 3. The draw down rate then slows down. In fact, it takes another three years

to hit a 90% rate. By year 10, the end of its expected life, the average fund is 93.6% invested. While

some funds remain in operation beyond year 10, there are no further draw downs.

11

Though not shown in the figure, there is wide variation in the speed with which funds draw down

committed capital. For instance, some funds draw it down in year 1, while others take as long as ten

years to invest 80% or more of their commitments. Adjusting for the fact that many of the more recent

funds are right-censored, in that they drop out of our sample before they are fully invested, the average

(median) fund takes 11.7 (11) quarters to invest 80% or more of its commitments.

On the flip side of the draw down decision, following liquidity events (such as an IPO), capital is

returned to investors in the form of cash distributions or stock distributions. (Private equity funds

typically have covenants restricting reinvestment of capital gains; see Gompers and Lerner (1996).)

Table 2B shows how much of the invested and committed capital was returned to investors by the

earlier of the end of our sampling period or a fund’s liquidation date. The average fund distributed

106.8% of drawn-down capital and 94.3% of committed capital. Again, this understates cash flows as

recent funds have yet to exit many of their portfolio holdings. The 73 funds raised between 1981 and

1993 returned 2.59 times invested capital and 2.45 times committed capital, on average.

Figure 1 documents the rate at which capital returns and capital gains are distributed to investors

over the life of the average fund. Several observations are in order. First, as one might expect,

distributions are rare in the early fund years. For example, by the end of year 3, only 12.9% of total

committed capital has been distributed on average, respectively. Second, it takes around seven years

for committed capital to be returned, so much of the “capital gain” is generated from year 7 onwards.

By year 10, the average fund has distributed 1.93 times its committed capital. Third, some funds

experience further capital distributions beyond year 10, either because they remain in operation or due

to a prolonged liquidation phase. By the time they are eventually liquidated, these funds have returned

2.62 times their committed capital, on average. It is important to note, therefore, that there can be

considerable payoff from private equity investments even after 10 years of operation.

These results have important implications for measuring performance and the liquidity of investing

in a PEF. Specifically, draw downs (cash outflows) and distributions (cash inflows) are the raw inputs

when assessing fund performance, but there is another ingredient: the time profile of cash flows. The

later the cash outflows, and the sooner the cash inflows, the better is a fund’s performance. Figure 1

shows that these cash flows occur throughout the life of the fund and thus must be taken into account

at the time they occur when calculating a fund’s return. This is a useful stylized fact therefore that

should be incorporated by the literature on PEF performance.

4.2 The Determinants of Draw Downs

To shed light on the determinants of how quickly a fund invests its capital, we model the time-to-

12

fully-invested as ln(ti) = βX + ln(εi), where the error εi is assumed to follow the exponential

distribution with mean β0, the constant. This is a standard accelerated-time-to-failure model, which is

perhaps more familiar when rewritten as a proportional-hazard duration model. One advantage of

failure models is that the likelihood function has no problem correcting for the right-censoring inherent

in the data (Kalbfleisch and Prentice (1980)).7 Thus, we estimate the model using all sample funds

raised between 1981 and 2001, including those that drop out of the sample before becoming fully

invested. (Our results are qualitatively unaffected if we restrict the sample to the mature funds raised

between 1981 and 1993, which are not subject to right-censoring.)

The model outlined in Section 2 suggests that time-to-fully-invested varies with the (time-varying)

availability of investment opportunities and competition for such investment opportunities, such that

funds invest their capital more rapidly when technological and other shocks increase the availability of

promising ventures and when they face less competition for deal flow. We also allow for potential

differences between venture and buyout funds, first-time and follow-on funds, and by fund size, and

control for changes in the cost of capital.

As our proxy for the unobserved availability of investment opportunities faced by a buyout

(venture) fund in our sample, we use the quarterly log number of companies in a buyout (venture)

fund’s industry of specialization that receive buyout (venture) funding according to Venture

Economics. Funds’ industry specializations are as defined in Section 3.4. For instance, an increase in

the number of “Biotechnology” companies being funded is assumed to signal an improvement in

biotech investment opportunities. This variable is time-varying: when investment opportunities (as

proxied by our variable) change over the life of a sample fund, the fund’s managers can respond by

accelerating or decelerating the rate at which they invest. Given the framework outlined in Section 2,

the time-varying nature of our proxy for investment opportunities is crucial, as we are interested in

how investment behavior responds to changes in the demand for private equity when the supply of

private equity is sticky in the short run.

We augment this proxy for investment opportunities with a “bubble dummy” that equals one

during the heyday of the new-economy boom (1999Q1-2000Q2), on the assumption that investment

opportunities were more abundant in those years. This too is a time-varying covariate: over the fund’s

7 In the absence of censoring, the likelihood of the data is simply the product of the conditional densities f(ti|β,xi) for all observations i. For a censored observation, the time at which “failure” occurs is unknown, as failure occurs after the end of the observation period, T. All that is known is that failure hasn’t yet occurred as of time T. The appropriate contribution to the likelihood function of a censored observation is therefore the probability of not having failed prior to T.

13

life, it equals one only in 1999Q1-2000Q2.8

To proxy for the degree of competition faced by a buyout (venture) fund in our sample, we

construct three variables. The first measures how much financial “fire power” the fund’s most direct

competitors have access to, and is defined as the amount of capital committed to buyout (venture)

funds in the year the sample buyout (venture) fund was raised, in log dollars of 1996 purchasing

power. This definition assumes that (say) a 1990 vintage fund competes primarily with other funds of

that vintage. (Our results are qualitatively unchanged if we widen the window to include capital

committed in the year before and after the fund’s vintage year.) Note that this variable is not time-

varying and is similar to the proxy used by Gompers and Lerner (2000).

Our second variable, “aggregate per-industry disbursements”, attempts to provide a proxy for

competition for individual deals. It is defined as the real aggregate amount of capital invested by all

Venture Economics funds in companies that fall within the sample fund’s industry specialization. For

instance, a buyout fund specializing in cable company acquisitions (VE group “Communications and

Media”) is treated as facing competition for deal flow from other funds investing in “Communications

and Media” companies. We measure aggregate per-industry disbursements during a fund’s first three

years, as Figure 1 shows that this is when funds invest most actively. We expect that funds take longer

to invest their capital, the more other funds invest in their industry of interest. Note the difference

between this and our first proxy for competition: while the first proxy is a measure of the fire power

available to a fund’s competitors, the second is a measure of how much capital competitors are actually

investing in the fund’s industry of interest.

The third measure of competition seeks to control for the fact that the private equity market clearly

grew and developed over the past two decades, becoming more competitive in the sense of greater

market acceptance of the PEF business model and thus, presumably, lower barriers to entry for new

funds. This suggests a time trend in the degree of competition existing managers face, with funds

raised earlier facing less competition than those raised in later years. To capture this, we include a

trend variable that equals the inverse of the square root of the fund’s vintage year, scaled such that

1981 equals 1 and later years have lower values. Given this definition, we expect a negative coefficient

for the trend variable.

We include a number of controls for fund characteristics, specifically the size of the fund (in log

real dollars), the type of fund (buyout versus venture), and the fund sequence (first-time versus follow-

8 Gompers and Lerner (2000) use price/earnings and market/book ratios of public firms in CRSP and Compustat to control for industry-specific investment opportunities among private firms, but find neither to be statistically significant.

14

on). We also control for changes in the cost of capital, using two measures: the yield on corporate

bonds (using Moody’s BAA bond index estimated quarterly in March, June, September, and

December), and the quarterly return on the Nasdaq Composite Index. Both are time-varying over the

life of a sample fund.

Table 3 reports the maximum-likelihood estimation results for three different cut-offs of “fully-

invested” (more than 70%, 80%, or 90% of committed capital).9 The results are qualitatively similar in

each case. The table also reports models estimated separately for buyout and venture funds using the

80% cut-off. (Qualitatively similar results, not shown, obtain for the 70% and 90% cut-offs). The

model χ2 statistics are large and highly significant in the three pooled models as well as in the buyout-

only and venture-only specifications, indicating good overall fit. The pseudo R2 suggest that our

models capture around a quarter of the variation in draw down rates.

15

equilibrium implications of Inderst and Mueller (2003), in the sense that PEF managers become more

cautious when competition for deal flow intensifies. The coefficient estimated for the variable

capturing the time trend in the evolution of the private equity market is negative and significant,

suggesting that funds raised earlier in the period, when the PEF market was less developed, were

invested significantly faster. (Note that this finding is not driven by the fact that many newer funds

drop out of our sample before becoming fully invested, as we have corrected for right-censoring.) To

illustrate the economic significance of these effects, we consider one-standard deviation increases in

the amount of competing PEF capital raised and disbursed (measured in log real dollars). All else

equal, these increase the time-to-fully-invested from the average of 11.7 quarters to 24.7 and 17.7

quarters, respectively.

Among fund characteristics, we find that venture funds take significantly longer to invest than

buyout funds. We find no evidence that fund sequence number or fund size affect the investment rate.

Increases in the cost of capital, as measured by the corporate bond yield, serve to reduce draw-down

rates, indicating that funds invest more slowly as debt becomes more expensive. The effect is fairly

large economically: all else equal, a one-standard deviation increase in bond yields would increase

time-to-fully-invested from 11.7 to 29.1 quarters. Conditions in the public equity markets, on the other

hand, do not influence investment behavior, in view of the insignificant coefficient estimated for the

return on the Nasdaq Composite Index. Of course, these conditions are above and beyond those

already captured by the investment opportunities and competition in the PEF market.

When we estimate the model separately for buyout and venture funds, we find similar results with

one exception: aggregate per-industry disbursement, our second proxy for competition for deal flow,

only has a significant effect on the draw down behavior of buyout funds. Venture funds are relatively

more sensitive to our first competition proxy, the amount of money available to funds raised in the

same vintage year. Note that changes in bond yields affect the investment behavior of both VC and

buyout funds. Conversations with the Limited Partner suggest this effect either captures the fact that

many venture funds in the sample specialize in “growth equity”, which more likely involves debt

financing, or “style drift” blurring the distinction between venture and buyout funds in the sample.

In conclusion, these duration models provide supporting evidence for our hypothesis that fund

behavior regarding investment decisions is a function of shocks to the availability of investment

opportunities, lags in the PEF market’s ability to respond to such shocks, and changes in the degree of

competition for deal flow.

16

4.3. The Determinants of Capital Returns

Barry, Muscarella, Peavy, and Vetsuypens (1990), Lerner (1994), Gompers (1996), Brav and

Gompers (2003), Black and Gilson (1998) and others have studied how private equity funds exit their

portfolio companies. A key finding from this literature is that PEFs act strategically in their exit

decisions, especially with respect to current market conditions and their need to build reputations. Our

model of the PEF market suggests that competition and investment opportunities may also affect exit

decisions. In this section, we model how PEFs exit their investments when the PEF market adjusts to

changes in investment opportunities with a lag and the degree of competition for deal flow varies over

time. This analysis complements Gompers and Lerner’s (2000) “money chasing deals” analysis, which

shows that current valuations of portfolio companies are high when there is significant competition for

deals.

To shed light on the determinants of how quickly a fund returns capital to its investors, we model

the log of time (in quarters) between a fund being created and it returning at least M times the

committed capital to the limited partners, using again accelerated-time-to-failure models.10 We

experiment with different cut-offs for M, and report estimation results for M=1x, 1.5x, and 2x capital.

Adjusting for the fact that many of the more recent funds are right-censored, in that they drop out of

our sample before they have had a chance to return their committed capital, the average (median) fund

has returned 1x capital after 18.8 (18) quarters, with correspondingly longer periods for the higher cut-

off points.

What determines capital returns? Having invested their capital, we expect funds to exit their

portfolio companies (and so return capital to their limited partners) more rapidly, the more public-

market investors are willing to pay for them. Their willingness to pay should increase in the investment

opportunities available to companies in an industry. Thus we expect faster capital returns, the better are

investment opportunities. As in Section 4.2, we proxy for investment opportunities using the log

number of companies receiving financing in the fund’s chosen industry of specialization. We also

include a dummy equaling 1 during calendar years 1999 and 2000, on the assumption that the exit

market was particularly favorable in those two years.

Funds facing tougher competition for deal flow find it harder to invest – as shown in Section 4.2 –

10 We model return of committed rather than invested capital because PEFs do not invest their committed capital instantly. To see the difference, consider the example of a fund that has drawn down 20% of commitments by year 2, distributes 25% of committed capital following an early “home run” in year 2, but takes until year 5 to invest all its committed capital and until year 9 to return it to investors. Time-to-return of “invested capital” would be two years, while time-to-return of committed capital would be nine years. The latter is more economically meaningful.

17

which implies that they will also take longer to return capital to their investors. We use the three

proxies for competition for deal flow introduced earlier: the total capital raised by other PEFs in the

fund’s vintage year (our measure of fire power), aggregate disbursements by other PEFs into the fund’s

main industry of interest (our measure of the amount of money chasing similar deals), and the time

trend variable.

Finally, we control possible differences across funds by including variables identifying venture and

first-time funds, respectively, and log fund size. We also control for the effects of variations over time

in capital market and exit market conditions. We identify and investigate four possible factors: (i) the

corporate bond yield as a measure of the cost of capital; (ii) the quarterly return on the Nasdaq

Composite Index, intended to capture the well-documented link between IPO activity and market

conditions (Loughran, Ritter, and Rydqvist (1994)); (iii) the quarterly number of private equity-backed

IPOs in the same broad VE industry, as a signal of how “hot” the IPO market is; and (iv) the quarterly

number of private equity-backed M&A deals in the same broad VE industry, as a signal of how “hot”

the M&A market is. Cheap debt, well-performing stock markets, receptive IPO markets and active

M&A markets should favor faster return of committed capital.

Table 4 reports maximum-likelihood estimates for the pooled sample using the three cut-offs and

separately for buyout and venture funds using the 1x cut-off (qualitatively similar results obtain for the

1.5x and 2x cut-offs). The model χ2 statistics are large and highly significant in all five models,

indicating good overall fit. The pseudo R2 suggest that our models capture around a half of the

variation in capital return decisions.

As predicted, funds return capital significantly faster, the more companies receive financing in a

fund’s industry of interest. This is true for all cut-offs and for venture funds and buyout funds

separately. Recall that we interpret an increase in the number of investments as an improvement in

opportunities and valuations. For example, if the outlook for optical switches improves, we would

expect more new ventures in the optical switches space to be funded, and at the same time existing

funds with investments in such companies should find it easier to exit them. To illustrate the economic

magnitude of the effect, a one-standard deviation increase in companies receiving financing in their

chosen industry of specialization would cut the time to returning 1x the committed capital from the

average of 18.8 to 4.2 quarters.

Competition for deal flow leads to slower capital returns, as conjectured. The effect is large

economically: a one-standard-deviation increase in the variable measuring the amount of capital

available to same-vintage year funds would delay the return of 1x committed capital by nearly two

18

years, from 18.8 to 25.8 quarters. How much competing funds actually invested in a sample fund’s

chosen industry has a positive effect on time-to-returning capital, but this is significant only in the

pooled 1x and venture-only specifications. The significantly negative coefficient estimated for the

trend variable shows that funds raised earlier returned their capital more rapidly, consistent with the

notion that earlier funds faced a less competitive PEF environment generally.

Fund characteristics are not generally significant, with one exception: venture funds are

significantly faster than buyout funds at returning their committed capital. Market conditions also play

a key role. Both buyout and venture funds return capital faster, the cheaper high-yield debt becomes,

while venture funds return capital faster, the higher are returns on the Nasdaq Composite Index. Both

effects are fairly large economically, with one-standard deviation changes in these variables leading to

reductions from 18.8 quarters to 8.9 and 14.0 quarters in the pooled sample, respectively. The climate

in the IPO market has no significant effect, but improved conditions in the M&A market (as measured

by an increase in the time-varying log number of M&A deals completed in a fund’s industry of

specialization) lead to a large reduction in the time to returning 1x the committed capital, from the

average of 18.8 to 11.2 quarters. This effect is concentrated among buyout funds.

In conclusion, these duration models provide supporting evidence for our hypothesis that fund

behavior regarding capital return decisions is a function of shocks to the availability of investment

opportunities, lags in the PEF market’s ability to respond to such shocks, and changes in the degree of

competition for deal flow, controlling for market conditions.

5. Portfolio Company-level Analysis of Private Equity Funds

5.1 The Determinants of Individual Exit Decisions

Having shown that fund-level decisions regarding capital returns are driven, at least in part, by

investment opportunities and competition considerations, we now analyze fund behavior regarding

individual exit decisions at the portfolio-company level. Thus the unit of observation in this section is a

portfolio company rather than a fund. This provides a micro-level foundation for the analysis in the

previous section.

Specifically, to see what determines how quickly a fund exits its investments, we model the log of

time (in quarters) between a fund investing in a given portfolio company and the fund distributing cash

or stock to its limited partners after exiting the investment (typically via an IPO or a sale). Note that

when a fund exits an investment in several stages, we use the first transaction date. Adjusting for the

fact that many of the more recent funds are right-censored and that failing investments are never

exited, the average (median) holding period is 14.4 (12) quarters, with a range from one to 62 quarters.

19

As before, we estimate standard accelerated-time-to-failure models using maximum likelihood,

first pooling all investments and then separately for the portfolio companies of buyout and venture

funds. We treat investments that are not exited by the earlier of the end of our sample period or the

tenth anniversary of a fund’s raising as right-censored, with corresponding modifications to the log-

likelihood function. Therefore, we estimate the model using the investments of all sample funds raised

between 1981 and 2001. (Our results are qualitatively unaffected if we restrict the sample investments

to those made by the mature funds raised in 1981-1993, which are not subject to right-censoring.)

We conjecture that holding periods are shorter (investments are exited faster), the better the

investment environment in terms of available opportunities and the less competition the fund faces. We

use the same proxies for these determinants as before, except that we measure per-industry

disbursements by other funds in the quarter an investment was actually undertaken (as opposed to

during a fund’s first three years of existence). This more directly captures the degree of potential

competition for the individual investment. The intuition for this proxy is that, holding the number of

companies funded constant (i.e., investment opportunities), an increase in the amount of money the

companies receive corresponds to an increase in valuations, all else equal, which is a measure of

“money chasing deals”.

We also control for three fund characteristics and two investment characteristics. First, venture

funds may have longer holding periods than buyout funds to the extent that they invest in less mature

companies that require more “value-added” input by the venture capitalists. Second, larger funds may

have a comparative advantage in seizing favorable exit opportunities, perhaps by virtue of having

stronger relationships with top IPO underwriters. Third, Gompers (1996) identifies a fund’s sequence

number as a potentially important factor in the exit decision, with first-time funds having an incentive

to take companies public too early (“grandstanding”). Fourth, larger investments potentially have more

of an impact on a fund’s profitability and IRR, and so may be exited sooner all else equal. Fifth, the

fund year (counted from 1 to 10) in which an investment was made may influence holding periods to

the extent that investments undertaken late in a fund’s life need to be unwound when the fund’s limited

partnership agreement expires (typically after ten years).

The final, and possibly most important, set of controls relates to market conditions. We use the

same four variables as in Table 4: the yield on high-yield corporate bonds, the quarterly return on the

Nasdaq Composite Index, and conditions in the two primary exit markets: the IPO market and the

M&A market. The latter two are conditioned on a sample fund’s industry of specialization. For all four

variables, we expect that PEFs exit their investments faster, the better the market condition (i.e., low

20

debt cost, high returns, strong IPO market, and active M&A market). Unlike the fund and industry

characteristics, market conditions change between the time an investment is undertaken and it is exited.

Table 5 reports the maximum-likelihood estimation results. The model χ2 statistics are large and

highly significant in the pooled model as well as in the buyout-only and venture-only specifications,

and the pseudo-R2 indicate that our models capture a good deal of the variation in holding periods.

Across all three models, improvements in the investment environment, as captured by our proxy,

lead to significantly faster exits. In the pooled specification, a one-standard-deviation increase in the

number of companies being funded reduces the holding period for the average portfolio company by

one year, from 14.4 to 10.3 quarters, holding all other covariates in the pooled model constant.

Consistent with investment opportunities being more plentiful during the heyday of the new-economy

boom, the significantly negative coefficient estimated for the bubble dummy shows that holding

periods dropped substantially in 1999-2000.

Competition for deal flow plays an important role in determining a PEF’s exit decisions: holding

periods are significantly longer when a PEF faces greater competition, as captured by increases in the

amount of capital available to a fund’s direct competitors and the aggregate amount of money chasing

deals in the same industry. As for the economic effects, one-standard-deviation increases in the amount

of capital available to same-vintage-year funds and of capital chasing similar deals increase the

average holding period in the pooled model from 14.4 to 19.9 and 19.0 quarters, respectively. In the

sub-sample models, we find the same signs and roughly the same economic effects. Finally, the trend

variable measuring the evolution of the PEF market has the expected negative coefficient, suggesting

that funds raised earlier exited their investments faster, ceteris paribus. Again, note that this is not

driven by right-censoring.

Among the controls, the most consistent effect comes from investment size: larger holdings are

exited significantly faster, with a one-standard-deviation increase accelerating the exit decision by two

and a half quarters in the pooled sample. Since shorter holding periods imply higher IRRs, ceteris

paribus, this suggests that PEF managers focus their attention on those investments that have the

largest impact on their fund returns. We also find that venture funds hold their investments

significantly longer than do buyout funds, consistent with venture investments requiring more time to

mature. Larger venture funds hold their investments significantly longer, which at first sight is odd:

larger funds are more likely to hold later-stage investments which ceteris paribus should be exited

faster. A possible explanation is that larger funds devote less time and attention to each portfolio

company (assuming VC skills are scarce) which in turn “mature” less quickly. We don’t find any

21

significant difference between first-time and follow-on funds, not even among venture funds, despite

their incentive to “grandstand”.

As one might expect, market conditions are an important determinant of the exit decision. For

example, as high-yield debt becomes more expensive, exits are delayed. This effect is present both for

buyout funds, which naturally are tied heavily to the leverage market, and for venture funds, be it

because they focus on “growth equity” or due to “style drift”. Economically, the effect is large: in the

pooled model, a one-standard deviation increase in bond yields lengthens mean holding periods from

14.4 to 21.9 quarters. An upturn in IPO activity also accelerates exits, especially among buyout funds,

with mean holding periods falling from 14.4 to 11.0 quarters following a one standard deviation

increase in log IPO volume.11 This provides complementary evidence to Barry, Muscarella, Peavy, and

Vetsuypens (1990) and Lerner (1994) who document in a variety of ways that venture capitalists have

market timing ability when taking companies public. Our result shows that the length of time they hold

an investment is a direct function of the IPO market climate. The return on the Nasdaq Composite

Index and conditions in the M&A market do not influence holding periods.

5.2 Hit Rates and Investment Multiples

Our data enable us to calculate investment-level returns for each portfolio company. We define

investment multiples as |cash inflows/invested capital|, ignoring the time value of money.12 Multiples

range from zero to ∞, with values between zero and one indicating capital losses. Funds of more recent

vintages still hold many unexited investments as of the end of our sample period, for which multiples

are necessarily zero. We therefore focus (for now) on the 73 mature funds raised between 1981 and

1993. Between them, these held 1,489 investments.

Among mature funds, the average portfolio company generated a multiple of 1.625. The

distribution is significantly right-skewed: 54.9% of investments were written off (i.e. zero multiples),

14% lost money (i.e. multiples less than one), 11.8% were “one-baggers” (i.e. multiples between one

and two), 6.3% were “two-baggers” (i.e. multiples between two and three), and the remaining 12.9%

were at least “three-baggers” (i.e. multiples of three or more).13 Broken up by fund type, we find that

11 If we use market-wide IPO activity rather than conditioning IPO volume on Venture Economics industries, the effect becomes larger in economic magnitude, without affecting the other results. 12 For successful investments, we know the dates of cash inflows and so could easily discount the flows. This is not the case for unsuccessful investments, as there are no inflows. Thus we ignore the time value of money to avoid making arbitrary assumptions about the length of the holding periods of investments that are written off. This also implies that we cannot compute annualized returns for unsuccessful investments. 13 Our distribution of multiples is broadly consistent with Cochrane’s.(2003) analysis of the fate of a broad sample of venture capital investments. Cochrane finds that 21.4% of the sample companies go public and 20.4% are acquired, with the remaining 58.2% classified as out of business or “still private”.

22

complete write-offs are much more common among venture funds (75.3%) than among buyout funds

(37.8%), though buyout funds have many more losses (21.3% vs. 5.4%). This indicates that buyout

investments – unlike venture investments – have some salvageable value even when they fail. Overall,

the portfolio companies of buyout funds have somewhat larger average multiples (1.69 vs. 1.55),

though the difference is not significant.14

What determines whether a particular portfolio company performs well or poorly? Obviously,

performance will have a large idiosyncratic component, driven by technology risk, the quality of

execution, market acceptance, competitors’ reactions and so on. However, the framework proposed in

this paper suggests that performance should also systematically be affected by changes in

entrepreneurs’ demand for capital, funds’ ability to react by supplying capital at short notice, and

competition for deal flow. Specifically, an improvement in investment opportunities should lead to

higher returns for an existing fund that can satisfy the demand for capital before new PEFs enter the

market. Conversely, tougher competition for deal flow should, all else equal, reduce performance.

We test these hypotheses by regressing investment multiples on proxies for investment

opportunities and competition for deal flow, controlling for fund characteristics (venture vs. buyout,

fund sequence number, and fund size), investment characteristics (size of investment and fund year in

which it was undertaken), and market conditions (the corporate bond yield at the time of investment).15

As a crude filter for idiosyncratic factors that may influence performance, we also include industry

fixed effects based on the six broad VE industry groups. Note that as in Section 5.1, the unit of

observation in this analysis is a portfolio company rather than a fund. We pool venture-backed and

buyout investments; results for each sub-sample are similar and are not reported. Standard errors are

adjusted for clustering on fund name (that is, investments undertaken by the same fund are not

assumed to be independent).

The regressions are estimated using ordinary least-squares16 which – in contrast to the duration

models estimated so far – provides no easy way to correct for right-censoring: funds raised more

recently are less likely to have reached the point where investments can be exited, so their portfolio

14 These results pool all portfolio companies from our sample. This does not take into account that there is variation across funds in the number of investments held. However, the corresponding results averaged by fund are not qualitatively different. 15 Another possible determinant of investment multiples is the market climate at the time of exit. Obviously, this is available only for exited investments and so reduces the sample size substantially. Conditional upon exit, we find that conditions in the IPO market (but not in the M&A market) have a significant effect on multiples, using the proxies introduced earlier (results not reported). 16 We obtain qualitatively identical results in probits of the likelihood that the multiple will exceed 1, 2, or 3.

23

companies are more likely to have zero multiples.17 Therefore, we estimate the model over different

samples, beginning with the investments held by funds raised in 1981-1993 (the mature funds in our

dataset) and adding later vintage years one by one. As more vintages are added, sample size grows but

the risk of right-censoring bias increases.

Table 6 reports the estimation results. Adjusted R2 range from 2.3% to 3.9%, suggesting that much

of the variation in performance is due to factors that we have not controlled for, including presumably

idiosyncratic factors. Improvements in investment opportunities have the predicted positive effect on

multiples, and this is significant across all regressions. Among mature funds, for example, a one-

standard deviation increase in the log number of companies funded in the same industry at the time a

sample company received its first investment increases the average multiple from 1.625 to 2.231,

holding all other covariates at their sample means. Interestingly, we also find that investments made

during the heyday of the new-economy boom in 1999-2000 subsequently had lower multiples, though

this is significant only in some of the specifications. Note that the positive relation between investment

opportunities and investment multiples is consistent with the framework of Section 2 and the results of

Section 5.1. That is, if existing PEFs are able to take advantage of “sticky” capital markets, then their

returns on investment should reflect this competitive advantage.

Tougher competition for deal flow, on the other hand, reduces multiples as conjectured: the more

money is available to a fund’s main competitors, and the more money is invested in the same industry,

the lower are multiples. To illustrate, a one-standard deviation increase in the amount invested in other

companies in the industry reduces multiples to 0.961 on average, using the estimates for vintage years

1981-1993. The interpretation of this result is that, holding investment opportunities fixed (as

measured by number of companies funded), the increase in money chasing deals reduces the

investment multiple by almost 40%. Our finding also complements Gompers and Lerner’s (2000)

analysis of the positive impact of capital inflows into venture funds on the “pre-money valuations” of

investments such funds undertake. Arguably, our results based on multiples provide a clearer picture of

the negative effect of competition on returns, since Gompers and Lerner do not know what fraction of

the equity VCs acquire in return for their investments.

Taken together, our evidence of a relation between multiples and both investment opportunities and

17 Alternatively, one might consider estimating censored regressions (such as a Tobit). This is problematic for two reasons. First, we face the practical problem of which investments in our data have zero multiples because they have been written off (so their true multiple is indeed zero), and which have zero multiples because we don’t observe long enough for them to pay off (right-censoring). Second, censored regressions (unlike OLS) are not robust to departures from the assumption that the underlying distribution is normal (see Goldberger (1983)). Normality is not a good description of the distribution of investment multiples.

24

competition strongly support the central hypotheses proposed in this paper. Note that Tables 3 through

5 show that private equity fund managers time their investment and exit decisions in response to

competitive conditions in the PEF market. A corollary of this is that the PEFs’ actions should be

reflected in the success rate of these investments. Table 6 demonstrates that this is the case.

6. Final Remarks

What factors explain the investment behavior of private equity fund managers? This paper

proposes a framework based on an imperfectly competitive market for private equity in which demand

for private equity varies over time and the supply of private equity is “sticky” in the short run.

Increases in demand can, in the short run, only be met by existing funds which accelerate their

investment flows and earn excess returns. Increases in supply lead to tougher competition for deal

flow, and private equity fund managers respond by cutting their investment spending. Supply increases

possibly indicate overheating accompanied by poorer performance (e.g., Kaplan and Stein (1993) and

Gompers and Lerner (2000)).

Using a unique dataset of private equity funds over the last two decades, we document evidence

consistent with this framework by estimating the determinants of the draw down and exit decisions of

funds’ investments throughout their life. Controlling for fund characteristics and market conditions, we

show that the competitive environment facing fund managers plays an important role in how they

manage their investments. During periods in which investment opportunities are good, existing funds

invest their capital and exit their investments more quickly, taking advantage of the favorable business

climate. This tends to lead to better returns on their investments. In contrast, when facing greater

competition from other private equity funds, fund managers draw down their capital more slowly and

hold their investments for longer periods of time. Returns on investment undertaken when competition

was tougher are ultimately significantly lower.

The model of the private equity market described in this paper has implications for the literature on

fund performance. Conditioning on PEF compensation being homogenous across funds, investors with

access to funds that are in a position to take advantage of the stickiness of private equity capital should

earn excess expected returns. Remaining investors earn normal risk-adjusted rates of return. The

exception, however, is the set of investors who provide capital during overheated environments in

which potentially “too much money chases deals”. These investors, of course, earn poor returns.

Evidence presented in Table 6 supports this view at the individual investment level. Moreover, this

model and the investment behavior of fund managers documented here coincide with the recent

literature that provides evidence of the determinants of private equity fund performance (see, for

25

example, Kaplan and Stein (1993), Gompers and Lerner (2000), Kaplan and Schoar (2003) and Jones

and Rhodes-Kropf (2003), among others).

But there are many questions that remain unanswered. For instance, does the cross-sectional and

time-series distribution of PEF returns imply rational behavior on the part of investors? At a general

level, this depends on the investor’s ability to predict the current level of competition for PEF capital

and forecast future periods of investment opportunities. Consider Kaplan and Schoar’s (2003) result

that returns from follow-on funds are persistent and exceed those of first-time funds (also see our Table

6 for similar evidence at the individual investment level). Investors may be acting rationally by

investing in first-time funds to the extent that it provides them an option (perhaps not available to all

investors) of investing in a follow-on fund if the PEF is successful. To fully address this issue, the

results in this paper suggest one possible factor, namely the degree of competition in the market

throughout the life of the fund. A complete answer, however, needs to incorporate the risk

characteristics of the fund as well as the premium for liquidity (if any) that may vary across funds.

26

References

Barry, Christopher B., Chris Muscarella, John W. Peavy III, and Michael R. Vetsuypens, 1990. “The role of venture capital in the creation of public companies: Evidence from the going public process.” Journal of Financial Economics 27, 447-471.

Bengtsson, Ola, Steven N. Kaplan, Frederic Martel, and Per Strömberg, 2002. “Investment screening and market conditions: Evidence from venture capital.” Unpublished working paper, University of Chicago.

Black, Bernard S., and Ronald J. Gilson, 1998. “Venture capital and the structure of capital markets: banks versus stock markets.” Journal of Financial Economics 47, 243-277.

Brav, Alon, and Paul A. Gompers, 2003. “The role of lock-ups in initial public offerings.” Review of Financial Studies 16, 1-29.

Cochrane, John, 2003. “The risk and return of venture capital.” Unpublished working paper, University of Chicago.

Goldberger, A.S., 1983. “Abnormal selection bias” in: S. Karlin, T. Amemiya, and L.A. Goodman (eds.), Studies in Econometrics, Time Series and Multivariate Statistics, 67-84. Academic Press.

Gompers, Paul A., 1995. “Optimal investment, monitoring, and the staging of venture capital.” Journal of Finance 50, 1461-1490.

Gompers, Paul A., 1996. “Grandstanding in the venture capital industry.” Journal of Financial Economics 43, 133-156.

Gompers, Paul A., and Josh Lerner, 1996. “The use of covenants: An analysis of venture partnership agreements.” Journal of Law and Economics 39, 463-498.

Gompers, Paul A., and Josh Lerner, 1998. “What drives fundraising?” Brookings Papers on Economic Activity: Microeconomics, 149-92.

Gompers, Paul A., and Josh Lerner, 1999a. “An analysis of compensation in the U.S. venture capital partnership.” Journal of Financial Economics 51, 3-44.

Gompers, Paul A., and Josh Lerner, 1999b. The Venture Capital Cycle. MIT Press.

Gompers, Paul A., and Josh Lerner, 2000. “Money chasing deals? The impact of fund inflows on private equity valuations.” Journal of Financial Economics 55, 281-325.

Gorman, Michael, and William A. Sahlman, 1989. “What do venture capitalists do?” Journal of Business Venturing 4, 231-248.

Hellmann, Thomas J., and Manju Puri, 2002. “Venture capital and the professionalization of start-up firms: Empirical evidence.” Journal of Finance 57, 169-197.

27

Inderst, Roman, and Holger Mueller, 2003. “The effect of capital market characteristics on the value of start-up firms.” Journal of Financial Economics, forthcoming

Jones, Charles M., and Matthew Rhodes-Kropf, 2002. “The price of diversifiable risk in venture capital and private equity.” Unpublished working paper, Columbia University.

Kalbfleisch, J.D., and R.L. Prentice, 1980. The Statistical Analysis of Failure Time Data. John Wiley & Sons, New York.

Kaplan, Steven N., and Antoinette Schoar, 2003. “Private equity returns: Persistence and capital flows.” Unpublished working paper, University of Chicago.

Kaplan, Steven N., and Jeremy Stein, 1993. “The evolution of buyout pricing and financial structure in the 1980s.” Quarterly Journal of Economics 108, 313-358.

Kaplan, Steven N., and Per Strömberg, 2003. “Financial contracting theory meets the real world: Evidence from venture capital contracts.” Review of Economic Studies 70, 281-315.

Lerner, Josh, 1994. “Venture capitalists and the decision to go public.” Journal of Financial Economics 35, 293-316.

Lerner, Josh, and Antoinette Schoar, 2003. “The illiquidity puzzle: Theory and evidence from private equity.” Journal of Financial Economics, forthcoming.

Loughran, Tim, Jay R. Ritter, and Kristian Rydqvist, 1994. “Initial public offerings: International insights.” Pacific Basin Finance Journal 2, 165-199.

Moskowitz, Toby, and Annette Vissing-Jørgensen, 2002. “The returns to entrepreneurial investment: A private equity premium puzzle?” American Economic Review 92, 745-778.

Palepu, Krishna, 1990. “Consequences of leveraged buyouts.” Journal of Financial Economics 27, 247-262.

Sahlman, William A., 1990. “The structure and governance of venture-capital organizations.” Journal of Financial Economics 27, 473-521.

Sahlman, William A., and Howard Stevenson, 1986. “Capital market myopia.” Journal of Business Venturing 1, 7-30.

Wermers, R., 1999. “Mutual fund herding and the impact on stock prices.” Journal of Finance 54, 581-622.

29

Table 1. Sample overview The sample consists of private equity and venture capital funds raised between 1981 and 2001 (the “vintage years”). To protect the identity of the Limited Partner, we have agreed not to disclose the number of newly raised funds the Limited Partner invested in after 1993. We refer to the 73 funds raised before 1993 as “mature” funds. VC funds are those identified as “Venture Capital” by Venture Economics. Most non-venture funds are flagged as “Buyout” (90.4%); the remainder are flagged as “Generalist Private Equity” (3.8%), “Mezzanine” (4.8%), and “Other Private Equity” (1%). Fund size is the capital committed by investors to a fund in all closings, as reported by Venture Economics and corrected by us where needed using partnership reports prepared by the fund managers. Total fund size is the aggregate amount raised by all sample funds. The “VE universe” refers to all funds raised in the relevant sample period according to Venture Economics that are headquartered in the same countries as our sample funds (the U.S. and certain countries in Europe and Latin America). Commitment is the Limited Partner’s capital commitment to the funds. Total commitment is the aggregate commitment by the Limited Partner. Mean commitment is equally weighted. All monetary numbers are in nominal U.S. dollars.

1981-2001 1981-1993

all fundsbuyout funds VC funds all funds

buyout funds VC funds

No. of funds * * * 73 54 19 Fund type % that are VC funds (by number) 24.9 26.0 % that are VC funds (by fund size) 14.8 11.8 Fund size ($m) Total 207,011 176,443 30,568 36,704 32,381 4,322 Mean * * * 502.8 599.7 227.5 Median 367.5 452.0 200.0 233.0 271.5 75.0 % of VE universe covered (by capital) 17.5 29.3 6.3 20.2 27.7 3.7 Commitment ($m) Total 5,459.4 4,772.0 687.5 1,107.0 1,020.8 86.2 Mean * * * 15.2 18.9 4.5 Median 10.0 12.5 5.0 7.0 10.0 3.2 Commitment/fund size (%) Mean 4.7 4.1 6.6 4.6 4.2 5.9 Median 3.3 3.2 3.7 3.7 3.4 5.6 Fund sequence number (as % of funds by number) first-time funds 28.9 25.0 28.8 34.8 31.5 42.1 second-time funds 20.6 18.8 third-time funds 11.6 8.7 later funds 39.0 37.7

Table 2A. Draw downs by vintage year Fund managers typically draw down the limited partners’ capital commitment not when the fund is raised but when they wish to invest in a portfolio company. The average fund in our sample has drawn down 67.32% of committed capital. However, this understates draw downs as the more recent funds in the sample are not yet fully invested. Therefore, we also report draw down schedules for the 73 funds raised between 1981 and 1993.

All funds Buyout funds VC funds

Vintage Year

No. of funds

Average draw downs /

committed capital

Fraction of funds that are 70% invested

Fraction of funds that are 80% invested

Average draw downs /

committed capital

Average draw downs /

committed capital

1981-2001 * 0.6732 0.556 0.495 0.6671 0.69161981-1993 73 0.9474 0.959 0.890 0.9466 0.9498

1981 1 0.9991 1.000 1.000 0.9991 n.a.1983 2 0.8973 1.000 0.500 1.0000 0.79471984 5 0.9688 1.000 1.000 1.0091 0.90851985 4 1.0112 1.000 1.000 1.0149 1.00001986 6 1.0003 1.000 1.000 1.0003 1.00001987 8 0.8654 0.750 0.625 0.8555 0.88191988 12 0.9780 1.000 1.000 0.9760 1.00001989 11 0.9517 1.000 0.909 0.9410 1.00001990 4 0.9217 1.000 0.750 0.8647 0.97871992 6 0.9027 0.833 0.833 0.8588 0.99041993 14 0.9462 1.000 0.929 0.9397 0.96271994 * 0.9313 0.938 0.875 0.9219 0.99691995 * 0.9101 1.000 0.923 0.8913 1.01331996 * 0.9016 0.944 0.889 0.8928 0.93211997 * 0.7632 0.618 0.441 0.6784 0.96681998 * 0.6511 0.400 0.400 0.6454 0.67861999 * 0.4119 0.100 0.025 0.3598 0.62012000 * 0.1906 0.000 0.000 0.1970 0.17852001 * 0.1831 0.000 0.000 n.a. 0.1831

Table 2B. Capital distributions by vintage year Funds are typically ten-year limited partnerships, with possible extensions by a few years subject to the limited partners’ approval. Following liquidity events (such as an IPO), capital is returned to the limited partners in the form of cash or stock distributions. In the latter case, the LP may either sell the stock directly or hold it as a public market investment. We record only stock distributions that are sold (as virtually all are in our sample). At the end of the fund’s life, the general partner “liquidates” the fund by selling all remaining assets and distributing the cash to the limited partners. The liquidation phase can potentially take a few years. The panel shows average cumulative distributions divided by invested and by committed capital for all funds raised between 1981 and 2001, and 1981 and 1993, and by vintage year.

All funds Buyout funds VC funds

Vintage Year

No. of funds

Averagedistributions /

capital invested

Average distributions /

committed capital

Average distributions /

capital invested

Average distributions /

capital invested

1981-2001 * 1.0683 0.9434 1.0307 1.18021981-1993 73 2.5913 2.4517 2.5639 2.6693

1981 1 3.2780 3.2751 3.2780 n.a.1983 2 3.2168 2.9249 3.5901 2.84351984 5 3.0794 2.9797 3.5046 2.44151985 4 5.1357 5.1416 5.7111 3.40951986 6 3.8571 3.8577 3.7980 4.15281987 8 2.6453 2.3634 2.8899 2.23781988 12 2.0259 1.9661 1.9999 2.31231989 11 2.6084 2.4332 2.3998 3.54691990 4 1.9637 1.7902 1.6966 2.23081992 6 1.8777 1.6396 2.1632 1.30671993 14 1.9346 1.7836 1.4851 3.05841994 * 1.3123 1.1882 1.4227 0.53941995 * 1.2377 1.1478 0.9561 2.78681996 * 0.8367 0.7804 0.7657 1.08531997 * 0.5130 0.4348 0.3942 0.79821998 * 0.5966 0.4377 0.4547 1.28201999 * 0.1995 0.0918 0.2107 0.15462000 * 0.1187 0.0130 0.0849 0.17572001 * 0.0001 0.0000 n.a. 0.0001

Table 3. The determinants of draw-down rates The dependent variable is the log of the time (in quarters) between a fund being raised and it having drawn down at least X% of its committed capital. We use three cutoffs for X: 70, 80 and 90%. The explanatory variables are listed in the table and discussed more fully in the text. We estimate accelerated-time-to-failure models using maximum likelihood estimators that are corrected for the right-censoring caused by funds leaving our sample before they are fully invested. We thus include all sample funds raised between 1981 and 2001 in the analysis. The error is assumed to have an exponential distribution with mean β0 (the constant). This model is identical to a proportional-hazard duration model, and coefficients can easily be converted into hazard ratios. The intercepts are not reported. We use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

All funds Buyout funds VC funds

Time to investing at least … 70% of committed

capital

80% of committed

capital

90% of committed

capital

80% of committed

capital

80% of committed

capital

time-

varying? Investment opportunities log quarterly per-industry no. of companies funded yes -0.797*** -0.942*** -0.770*** -0.895*** -0.762* 0.170 0.170 0.164 0.198 0.415

dummy=1 if in 1999Q1 to 2000Q2 yes -1.441*** -1.071*** -0.705*** -1.018*** -1.149**

0.235 0.235 0.250 0.273 0.529 Competition for deal flow log real fund inflows, same vintage year no 0.737*** 0.772*** 0.366* 0.877*** 1.185***

0.223 0.216 0.217 0.272 0.382

log aggregate per-industry disbursement in first 3 yrs no 0.289** 0.384*** 0.453*** 0.213* -0.232 0.124 0.123 0.126 0.110 0.326

trend = vintage year–½ (scaled to give 1981=1) no -12.552*** -9.294*** -8.482*** -14.503*** -8.433**

2.532 2.193 2.334 3.548 3.997 Fund characteristics dummy=1 if venture fund no 1.238*** 1.500*** 0.952*** 0.323 0.318 0.307

dummy=1 if first-time fund no 0.027 -0.009 -0.241 -0.261 0.370 0.194 0.206 0.215 0.224 0.456

log real fund size no -0.028 -0.075 -0.043 -0.012 -0.079 0.075 0.081 0.089 0.092 0.162 Cost of capital BAA corporate bond yield (in %) yes 0.773*** 0.721*** 0.589*** 0.945*** 0.634***

0.114 0.108 0.110 0.145 0.238

quarterly return on Nasdaq Comp. Index (in %) yes 0.009 -0.001 0.007 0.000 0.002 0.006 0.006 0.007 0.007 0.013 Pseudo-R2 22.5 % 25.3 % 21.7 % 25.3 % 29.7 % Likelihood ratio test: all coeff. = 0 (χ2) 113.4*** 120.8*** 97.4*** 91.0*** 34.4*** Number of observations (time at risk) 2,897 3,239 3,615 2,496 743

Table 4. The determinants of capital returns The dependent variable is the log of the time (in quarters) between a fund being raised and it having returned at least M times its committed capital. We use three cutoffs for M: 1x, 1.5x and 2x. The explanatory variables are listed in the table. We estimate accelerated-time-to-failure models using maximum likelihood estimators that are corrected for the right-censoring caused by funds leaving our sample before they are fully invested. We thus include all sample funds raised between 1981 and 2001 in the analysis. The error is assumed to have an exponential distribution with mean β0 (the constant). This model is identical to a proportional-hazard duration model, and coefficients can easily be converted into hazard ratios. Intercepts are not reported. The intercepts are not reported. We use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

All funds Buyout funds VC funds

Time to returning at least … committed capital 1x 1.5x 2x 1x 1x

time-

varying? Investment opportunities log quarterly per-industry no. of companies funded yes -1.558*** -0.639* -1.245*** -0.983*** -2.590***

0.301 0.368 0.425 0.370 0.564

dummy=1 if in 1999Q1 to 2000Q2 yes 0.221 -0.166 -0.454 -0.001 1.223 0.366 0.402 0.427 0.427 0.854 Competition for deal flow log real fund inflows, same vintage year no 1.215*** 1.121*** 0.930*** 0.921*** 1.601***

0.238 0.273 0.305 0.341 0.464

log aggregate per-industry disbursement in first 3 yrs no 0.560** 0.134 0.563 0.355 0.792* 0.249 0.309 0.362 0.315 0.472

trend = vintage year–½ (scaled to give 1981=1) no -8.933*** -8.533*** -9.408*** -11.998*** -7.246***

2.308 2.529 2.966 3.585 3.442 Fund characteristics dummy=1 if venture fund no 1.874*** 1.052** 1.124** 0.395 0.464 0.511

dummy=1 if first-time fund no 0.285 0.141 -0.031 0.227 0.365 0.256 0.309 0.351 0.295 0.535

log real fund size no 0.084 -0.048 -0.070 0.086 0.088 0.097 0.118 0.140 0.115 0.191 Market conditions BAA corporate bond yield (in %) yes 0.632*** 0.222 0.364 0.610*** 0.802**

0.170 0.213 0.272 0.202 0.398

quarterly return on Nasdaq Composite Index (%) yes -0.019** 0.001 0.003 -0.003 -0.047***

0.009 0.011 0.012 0.011 0.018

log quarterly per-industry number of IPOs yes 0.278 0.277 0.271 0.269 0.473 0.170 0.204 0.227 0.220 0.310

log quarterly per-industry number of M&A deals yes -0.579** -1.322*** -1.086*** -0.730** -0.078 0.272 0.368 0.419 0.328 0.528 Pseudo-R2 48.3 % 44.5 % 46.2 % 44.1 % 74.6 % Likelihood ratio test: all coeff. = 0 (χ2) 160.7*** 120.8*** 106.9*** 107.7*** 65.5*** Number of observations (time at risk) 5,207 5,647 6,011 3,926 1,281

Table 5. The determinants of holding periods The dependent variable is the log of the time (in quarters) between a fund investing in a given portfolio company and the fund distributing cash or stock to its LPs after exiting the investment (typically via an IPO or a sale). If the fund exits the investment in several stages, we use the first transaction date. Investments that are not exited by the earlier of the end of our sample period or the tenth anniversary of a fund’s creation are treated as right-censored, with appropriate modifications to the log-likelihood function. We thus include all sample funds raised between 1981 and 2001 in the analysis. The explanatory variables are listed in the table. We estimate accelerated time-to-failure models using maximum likelihood estimators. The error is assumed to have an exponential distribution with mean β0 (the constant). This model is identical to a proportional-hazard duration model, and coefficients can easily be converted into hazard ratios. The intercepts are not reported. Standard errors, shown in italics, are adjusted for clustering on fund name (that is, investments undertaken by the same fund are not assumed to be independent). We use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

time-

varying?

All funds Buyout funds VC funds

Investment opportunities log quarterly per-industry no. of companies funded yes -0.375*** -0.344*** -0.423** 0.105 0.122 0.216 dummy=1 if in 1999Q1 to 2000Q2 yes -0.624*** -0.584*** -0.842*** 0.125 0.141 0.261 Competition for deal flow log real fund inflows, same vintage year no 0.350** 0.335* 0.563** 0.150 0.201 0.236 log real per-industry disbursements at time of investment no 0.184*** 0.178*** 0.169** 0.031 0.033 0.086 trend = vintage year–½ (scaled to give 1981=1) no -6.477*** -6.144*** -6.976*** 1.466 2.159 1.746 Fund and investment characteristics dummy=1 if venture fund no 0.956*** 0.206 dummy=1 if first-time fund no -0.044 0.001 -0.038 0.152 0.180 0.257 log real fund size no 0.015 -0.024 0.141* 0.062 0.071 0.075 log real investment cost no -0.122*** -0.104** -0.232*** 0.037 0.042 0.065 fund year in which investment was made (1 to 10) no 0.054 0.040 0.109 0.036 0.037 0.096 Market conditions BAA corporate bond yield (in %) yes 0.524*** 0.517*** 0.527*** 0.078 0.095 0.130 quarterly return on Nasdaq Composite Index (%) yes 0.002 0.003 -0.002 0.003 0.003 0.007 log quarterly per-industry number of IPOs yes -0.220*** -0.290*** 0.021 0.052 0.060 0.108 log quarterly per-industry number of M&A deals yes -0.113 -0.159 -0.082 0.098 0.125 0.139

Pseudo-R2 14.7 % 13.8 % 18.6 % Likelihood ratio test: all coeff. = 0 (χ2) 333.6*** 290.9*** 118.6*** Number of observations (time at risk) 30,189 22,407 7,782

35

Table 6. Determinants of investment multiples The dependent variable is the multiple earned on each portfolio company, defined as |cash inflows/invested capital| and so ranging from 0 to ∞. The explanatory variables are listed in the table. We estimate ordinary least-square regressions over different samples, beginning with all funds raised between 1981 and 1993 (the mature funds in our dataset) and adding later vintage years one by one. Funds raised more recently are less likely to have reached the point where investments can be exited, so their portfolio companies are more likely to have zero multiples. Industry fixed effects, using broad Venture Economics industries, are included but not shown. Standard errors, shown in italics, are adjusted for clustering on fund name (that is, investments undertaken by the same fund are not assumed to be independent). We use ***, **, and * to denote significance at the 1%, 5%, and 10% level (two-sided), respectively.

Vintage years: 1981 1981 1981 1981 1981 1981 to 1993 to 1994 to 1995 to 1996 to 1997 to 1998

Investment opportunities log per-industry no. of companies funded 0.729** 0.512* 0.383 0.455* 0.426** 0.351* 0.315 0.274 0.258 0.239 0.208 0.191 dummy=1 if investment made in 1999Q1 to 2000Q2 -1.384 -0.827 -0.649 -0.841* -0.420 -0.453* 1.788 0.997 0.779 0.510 0.315 0.236 Competition for deal flow log real fund inflows, same vintage year -0.389 -0.601* -0.710** -0.697** -0.799*** -0.770***

0.348 0.311 0.298 0.288 0.223 0.197

log real per-industry disbursements at time of investment -0.577** -0.516** -0.448** -0.525*** -0.504*** -0.422***

0.227 0.205 0.198 0.183 0.161 0.149 trend = vintage year–½ (scaled to give 1981=1) -4.605 -4.294 -4.983 -3.254 -4.161 -3.685 3.815 3.459 3.348 3.120 2.702 2.503 Fund and investment characteristics dummy=1 if venture fund -0.862** -0.852** -0.784** -0.952*** -0.963*** -1.014***

0.395 0.367 0.355 0.334 0.275 0.247

dummy=1 if first-time fund -1.182*** -1.014*** -0.909*** -0.896*** -0.684*** -0.576***

0.374 0.320 0.296 0.279 0.235 0.217

log real fund size 0.021 0.058 0.106 0.082 0.074 0.111 0.134 0.117 0.112 0.102 0.087 0.076

log real investment cost -0.122* -0.114* -0.154** -0.144*** -0.121*** -0.228***

0.074 0.064 0.060 0.054 0.047 0.043

fund year in which investment was made (1 to 10) -0.311*** -0.216** -0.188* -0.163* -0.146** -0.152** 0.120 0.105 0.100 0.087 0.073 0.065 Market conditions at time of investment BAA corporate bond yield (in %) -0.103 -0.033 -0.018 -0.038 -0.025 -0.044 0.155 0.141 0.137 0.126 0.109 0.100 Adjusted R2 2.6 % 2.3 % 2.5 % 2.5 % 3.0 % 3.9 % No. of portfolio companies 1,489 1,727 1,946 2,236 2,703 3,199